Comparative Analysis of Anomaly Detection Algorithms for Development of the Unmanned Aerial Vehicle Power Plant Digital Twin

Abstract

This article considers the capabilities of machine learning algorithms in the problem of anomaly detection, analysis of their effectiveness in simultaneous processing of several telemetric data streams. A variant of constructing an algorithm using the Hierarchical Temporal Memory for the development of the power plant digital twin of a medium-altitude long-endurance unmanned aerial vehicle is considered. In terms of their basic characteristics (geometric, weight) such aerial vehicle is considered correspond to the light aircraft class. In particular, a modification of the algorithm for analyzing of telemetric information received from sensors of the on-board measurement system is considered.